RoboCade: Gamifying Robot Data Collection
Suvir Mirchandani, Mia Tang, Jiafei Duan, Jubayer Hamid, Chung Yeung Michael Cho, Dorsa Sadigh
AI summary
Problem
Scaling robot imitation learning is bottlenecked by the high cost and tedium of collecting demonstration data, which traditionally requires specialized hardware and trained operators. Existing remote teleoperation platforms fail to motivate broad participation due to their purely utilitarian design.
Approach
RoboCade transforms remote teleoperation into an engaging game by embedding feedback, leaderboards, narratives, and goal diversity into both the interface and manipulation tasks, enabling novice users to collect useful data remotely.
Key results
- Co-training with gamified data improved target task success rates by 16–56%
- Out-of-distribution policy performance increased by up to 20%
- Novice users rated the gamified platform 24–27% more enjoyable, intuitive, and motivating
- Validated across three distinct manipulation tasks: rearrangement, scanning, and insertion
Why it matters
Provides a scalable, incentive-free pathway for crowdsourcing high-quality robot demonstration data, lowering barriers for imitation learning research and deployment.
Abstract
Imitation learning from human demonstrations has become a dominant approach for training autonomous robot policies. However, collecting demonstration datasets is costly: it often requires access to robots and needs sustained effort in a tedious, long process. These factors limit the scale of data available for training policies. We aim to address this scalability challenge by involving a broader audience in a gamified data collection experi- ence that is both accessible and motivating. Specifically, we develop a gamified remote teleoperation platform, RoboCade, to engage general users in collecting data that is beneficial for downstream policy training. To do this, we embed gamification strategies into the design of the system interface and data collection tasks. In the system interface, we include components such as visual feedback, sound effects, goal visualizations, progress bars, leaderboards, and badges. We additionally propose principles for constructing gami- fied tasks that have overlapping structure with useful downstream target tasks. We instantiate RoboCade on three manipulation tasks—including spatial arrangement, scanning, and insertion. To illustrate the viability of gamified robot data collection, we collect a demonstration dataset through our platform, and show that co- training robot policies with this data can improve success rate on non-gamified target tasks (+16–56%). Further, we conduct a user study to validate that novice users find the gamified platform sig- nificantly more enjoyable than a standard non-gamified platform (+24%). These results highlight the promise of gamified data col- lection as a scalable, accessible, and engaging method for collecting demonstration data. Videos are available at robocade.github.io.